Using Quarterly GDP in Constructing a Monthly Coincident Index of Business Cycles
2004/10/26
研討會日期 : 2004-10-26
時間 : 15:00
主講人 : Prof. Roberto S. Mariano
地點 : B棟110室
演講者簡介 : Prof. Roberto S. Mariano為Ph.D. in Statistics，Stanford University(1970)。
現為Professor，Singapore Management University。
其主要研究領域為Theoretical and Applied Econometrics及Macroeconomic Modeling for Forecasting and Policy Analysis。
演講摘要 : The Stock-Watson coincident index and its subsequent extensions assume a static linear one-factor model for the component indicators – four indicators in the case of the U.S. This paper explores the possibility that observable quarterly GDP contains additional information that can be utilized in improving the monthly coincident index. We define a coincident index of business cycles as an estimate of latent monthly real GDP and estimate alternative models of the joint behavior of latent monthly real GDP and coincident indicators using the observable mixed-frequency series. Complications that arise from the missing observations on monthly GDP are addressed through a state-space formulation of the models. Dynamic factor models as well as vector autoregressive (VAR) models are considered. Maximum likelihood estimation is implemented through an EM algorithm. The associated smoothed estimate of latent monthly real GDP is the proposed new coincident index.
